From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [140.186.70.92] (port=41085 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1Pugue-0005bz-01 for qemu-devel@nongnu.org; Wed, 02 Mar 2011 02:56:02 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Pugud-00055v-12 for qemu-devel@nongnu.org; Wed, 02 Mar 2011 02:56:00 -0500 Received: from mx1.redhat.com ([209.132.183.28]:46789) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Puguc-00055h-LT for qemu-devel@nongnu.org; Wed, 02 Mar 2011 02:55:58 -0500 Message-ID: <4D6DF866.4060804@redhat.com> Date: Wed, 02 Mar 2011 08:57:26 +0100 From: Kevin Wolf MIME-Version: 1.0 Subject: Re: [Qemu-devel] [RFC][PATCH 0/2] Allow cache settings for block devices to be changed at runtime. References: <20110228171956.05a84fb9@zephyr> <4D6BBB72.6040205@redhat.com> <20110301124254.GA29485@lst.de> <20110301124834.GA10953@stefanha-thinkpad.localdomain> <20110301125036.GA29715@lst.de> <4D6D4569.8070502@codemonkey.ws> In-Reply-To: <4D6D4569.8070502@codemonkey.ws> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Anthony Liguori Cc: qemu-devel , Ananth Narayan , Christoph Hellwig , Prerna Saxena , Stefan Hajnoczi Am 01.03.2011 20:13, schrieb Anthony Liguori: > On 03/01/2011 07:50 AM, Christoph Hellwig wrote: >> On Tue, Mar 01, 2011 at 12:48:34PM +0000, Stefan Hajnoczi wrote: >> >>> On Tue, Mar 01, 2011 at 01:42:54PM +0100, Christoph Hellwig wrote: >>> >>>> I have patches to do that, and to allow changing O_DIRECT via a monitor >>>> command, but to toggle O_SYNC via fcntl I first need to get a kernel >>>> patch in as that's currently not allowed to be changed at runtime. >>>> >>> Great it sounds like you have already implemented the two cases (guest wce and >>> host O_DIRECT) that we're talking about. >>> >> At least in theory. And for Linux I can add setting/clearing of O_SYNC >> via fcntl easily, but what do we do for other hosts? > > To start with, we can just fail the command at the QMP level. O_SYNC isn't toggled from QMP but from the guest. Kevin